GOM Cooperative BLL Data Exploration

Data Exploration and Index of Abundance Comparison of Longline vs. Trawl Data

Author

Adam Kemberling

Published

May 27, 2025

Gulf of Maine Cooperative Bottom Longline Survey

The following section of text is from the NOAA technical memorandum NMFS-NE-249 which covers the design, purpose, and implementation of the Gulf of Maine Cooperative Research Longline Survey. The full report is stored with the data (acquired from David McElroy) in a folder within Box/Res_Data/GOM_Longline_Survey/GOM_Cooperative_BLL.

The following section contains two subsets of the memorandum’s abstract detailing the origin of the survey and some key details on methodology:

In 2013, the National Cooperative Research Program accepted and funded the Northeast Fisheries Science Center’s (NEFSC) Northeast Cooperative Research Branch to implement a bottom longline survey (LLS) in the western and central Gulf of Maine (GM). The objective of the proposed LLS was to increase sampling of several data-poor and depleted stocks specifically associated with rocky habitat, while also enhancing data collection for some data-rich stocks already seemingly well sampled on the bottom trawl survey (BTS). The LLS stratified random design was based on the NEFSC BTS stratification of depth and area, and this was further stratified by “rough” and “smooth” bottom type within the strata area to account for habitat. The survey area includes 6 BTS offshore strata: part of strata 28, 29, and 36 and all of strata 26, 27, and 37.

During 2014-2017, the survey sampled 45 station locations each spring (April-May) and fall (October-November), during the same seasonal period as the NEFSC research BTS, so that the LLS would overlap with the BTS not only spatially, but temporally as well, for comparability. Two commercial fishing vessels simultaneously conducted the survey. The survey longline gear is a 1 nautical mile/1,852 m groundline baited with frozen squid (Illex spp) on each of 1000 #12 semi-circle, easy-baiter hooks. A temperature/depth probe is secured on each anchor, as well as a current meter to measure the near-bottom velocities. The gear is set across slack tide for a minimum 2-hour soak time.

Data & Documentation

The package of data obtained from David McElroy contained the following datasets and documentation materials:

  1. Catch
  2. Length
  3. TD
  4. Data definitions
  5. Survey Design Documentation

I also requested shapefiles for the longline strata (which are slightly modified versions of trawl strata), which I also received.

Data Definitions

The following table provides a readout of the data dictionary for the longline dataset, with field names and their descriptions.

Data Dictionary
Gulf of Maine Cooperative Longline Survey
CRUISE6 Unique numerical code for each cruise. Specific to a vessel and season, where first 4 digits indicate year and next two indicate vessel and season (i.e., 44 and 46 = Mary Elizabeth, 45 and 47 = Tenacious II, 44 and 45 = Spring, 46 and 47 = Fall).
STATION Unique sequential order in which stations have been completed, unique to each cruise code.
YEAR NA
SEASON NA
ASSIGNED_STRATUM the set was assigned to the stratum where the set originated.
BOTTOM_TYPE bottom type sub-stratum preassigned to station
COMMON_NAME NA
SEX Code indicating sex of fish or invertebrate species. 0 = unknown. 1 = male. 2 = female.
LENWT total weight of catch for which lengths were taken
LENNUM total count of catch for which lengths were taken
CATCHWT total weight of catch
CATCHNUM total count of catch
COMMENTS NA
AUDIT_FLAG flag=Y indicates potential data issues worth looking at before using
BEGIN_SET_TIME Begin of gear deployment in UTC
END_SET_TIME End of gear deployment UTC
SOAK_DURATION_DH Soak duration in hrs
DECDEG_BEGLAT_SET NA
DECDEG_BEGLON_SET NA
DECDEG_ENDLAT_SET NA
DECDEG_ENDLON_SET NA
GEAR_EVALUATION see tech memo
GEAR_DAMAGE see tech memo
GEAR_INTERACTION see tech memo
INTERACTION_TYPE see tech memo
ENVIRONMENTAL_CONDITIONS see tech memo
DATA_ACQUISITION see tech memo
SET_NOTES NA
HAUL_NOTES NA
CATCH_NOTES NA
AUDIT_DONE audit complete

Difference Between Trawl & LL Strata

The cooperative longline survey uses modified strata for its sampling procedure which are shown below in bold.

Loading the Datasets

The three longline .csv files can be read in with any standard read function:

I’ve changed the field names to be lowercase using janitor::clean_names for personal preference, but otherwise everything is as-is.

The BLL_CATCH dataset contains the total abundance and weight of catch by species for each longline set. This is the primary dataset we would use for cpue estimates.

The BLL_LENGTH dataset contains individual length measurements for subsets of the overall catch by species. This dataset provides context on the size structure of the catch.

The TD_data dataset contains data from the CTD, and information about the stations.

Tidying Spatial Information

When doing some exploratory mapping of the BLL_CATCH data I noticed some instances in the longline data where the assigned_strata is four digits long (usually 2) and is actually two strata sandwiched together.

Following visual inspection of some maps, it did not appear to be the case that they were cases where the longline was set across strata, simply they had a label that was less specific to the specific strata, indicating instead the joint area.

This step checks that, and then uses sf to reassign to a specific strata.

There are ten instances where the longline sets do indeed cross the boundaries between strata:

For these I am just leaving them as NA (not a long-term solution). The starting location would probably make sense to use, this would be consistent with the trawl survey (pretty sure).

Most stations are okay though using their new strat assignments:

Mid-Atlantic Species

I was informed that there are 15 species of interest which the Mid-Atlantic Fisheries Management Council was interested in for this topic of shifting distributions across management jurisdictions:

There are five species that are present in both the list of species managed by the MAFMC and among the species sampled as part of the cooperative longline survey:

  1. Spiny dogfish
  2. Goosefish
  3. Tilefish
  4. Atlantic Mackerel
  5. Ocean Quahog

Atlantic Mackerel and Ocean Quahog’s are both special cases, where the species was sampled on one occassion.

Comparing Trawl vs. LL CPUE

For this section, I just want to do a quick side-by-side of two indices of species abundances from the two surveys.

As a simple entry-point, I am just going to get the area-stratified mean CPUE for the strata sampled by both gears.

The catch/effort is area-weighted by the size of each strata so that the cpue of smaller strata carries less weight than the cpue of a larger strata when aggregating them to a value for the full area.

Longline CPUE Indices

The mean abundance/set is first estimated each season for each of the longline strata. Then a weighted average of these estimates (weighted by the square km surface area of each strata) is taken to determine an area-wide cpue.

The area-wide values are then scaled against the long-term average to return an index value.

Trawl CPUE Indices

The trawl cpue indices are done similarly, they use the full trawl strata area for weighting (not the longline strata).

Combined Tracking

Get the indices into the same table for side-by-side display, with columns by season.

Hook Saturation

The nature of the longline survey means that there are cases where species to first arrive can prevent other species from being sampled.

As an interesting check, I wanted to see what fraction of hooks was occupied by a species each season/year vs. empty.

Capacity for Use as Indicator of Species Movements:

The cooperative longline survey has coverage in spring and fall from 2014-2024. This survey provides the only medium/long-term source of fisheries independent hook and line sampling for species associated with hard-bottom habitats which are difficult to sample with trawl gear.

To-date, there has been no sampling of black seabass, a species I would have expected to see in this dataset given their noted range expansions and their habitat associations.

There are three species which are monitored by the MAFMC that have good sampling by this survey, those are goosefish, spiny dogfish, and to a lesser degree tilefish.

This survey provides important insights for fisheries managers, but shows limited evidence for range expansion of Mid-Atlantic species at this time.

Splitting Trawl Survey into Regional Indices

Need something to convey what potential this dataset has for monitoring species range shifts.

Get the average cpue for a species within these areas:

Plotting Comparisons of CPUE

These are not scaled to the region and reflect what the relative densities are in the different areas:

Plotting Local Relative Abundance Changes

These are scaled to the long-term average of a region:

Comparing Area stratified abundance + biomass

This function should be able to process area stratified abundance and biomass from the trawl survey data.

It can be applied at the council scales by subsetting the catch data and the effort data down to those scales prior to running the function.

Returns the mean cpue (abundance and biomass), the stratified totals, and indices of relative abundance (scaled by long term mean and sd) for the whatever group stratification we are interested in.